Update README.md

README.md

@@ -1,3 +1,113 @@
 ---
-license: apache-2.0
+license: other
+base_model: Qwen/Qwen3.5-9B-Base
+tags:
+  - cobol
+  - mainframe
+  - code
+  - legacy-modernization
+  - qwen3_5
+  - lora
+language:
+  - en
+pipeline_tag: text-generation
 ---
+
+# FL-9B-4
+
+FL-9B-4 is a COBOL / mainframe code model fine-tuned from **Qwen/Qwen3.5-9B-Base**
+via supervised fine-tuning (SFT) on a curated COBOL instruction dataset. It targets
+legacy-code understanding, COBOL generation, and COBOL→Java translation.
+
+- **Base model:** Qwen/Qwen3.5-9B-Base (dense 9B, hybrid linear + full attention)
+- **Method:** LoRA SFT (assistant-only masking), ~3 epochs, bf16
+- **Domain:** COBOL, GnuCOBOL, mainframe knowledge, COBOL↔Java
+
+## Benchmark results
+
+All code benchmarks compile and execute generated programs against reference
+tests. Evaluated greedy (temperature 0), single sample per task.
+
+| Benchmark | Metric | FL-9B-4 |
+|---|---|---|
+| **COBOLEval** | pass@1 | **36.99%** |
+| | compile rate | 82.10% |
+| | test pass rate | 52.98% |
+| **COBOL-JavaTrans (C2J)** | pass@1 | **80.42%** |
+| | compile success rate (CSR) | 96.50% |
+| **MainframeBench** | MCQ accuracy | **71.26%** (1376/1931) |
+| | QA — Token F1 | 12.75% |
+| | QA — ROUGE-L | 10.29% |
+| | Summarization — Token F1 | 27.64% |
+| | Summarization — ROUGE-L | 20.25% |
+| **CobolCodeBench** | INSTRUCT compile rate | **47.83%** (22/46) |
+| | COMPLETE compile rate | 32.61% (15/46) |
+
+### Notes on evaluation
+
+The MainframeBench MCQ, CobolCodeBench INSTRUCT and COMPLETE numbers were produced
+after fixing harness-side generation limits (the default 16-token MCQ budget and
+2048-token code budget truncated answers, and single-format `cobc` invocation
+rejected valid programs written in a different column format). Fixed evaluation
+uses a larger generation budget and tries `variable`, `free` and `fixed` COBOL
+formats when compiling. Reported numbers reflect the model's actual capability,
+not the truncated defaults.
+
+The strongest results — COBOL→Java translation (80% pass@1) and COBOLEval
+(82% compile) — show the model reliably produces valid, working COBOL and
+translates legacy code into working Java.
+
+## Intended use
+
+- Translating legacy COBOL programs to Java
+- Completing and generating GnuCOBOL programs
+- Answering mainframe / COBOL knowledge questions
+- Assisting with legacy-code modernization workflows
+
+## Limitations
+
+- Open-ended QA and summarization scores (Token F1 / ROUGE-L) are modest; the
+  model is stronger at code generation and translation than at free-form prose.
+- COBOL generation quality varies with column-format conventions; generated code
+  may mix fixed and free formats.
+- Evaluated only on GnuCOBOL 3.2; behavior on other COBOL dialects/compilers
+  (IBM Enterprise COBOL, etc.) is untested.
+- Not evaluated for safety-critical or production mainframe deployment without
+  human review.
+
+## Training
+
+| Setting | Value |
+|---|---|
+| Base | Qwen/Qwen3.5-9B-Base |
+| Method | LoRA (r=32, α=64), assistant-only SFT |
+| Precision | bf16 |
+| Epochs | ~3 |
+| Sequence length | 8192 (packed) |
+| Hardware | 1× NVIDIA RTX PRO 6000 Blackwell (96 GB) |
+| Frameworks | Unsloth + Transformers |
+
+LoRA adapters were applied to attention projections, MLP projections, and the
+linear-attention (`in_proj_*`/`out_proj`) modules of the hybrid Qwen3.5
+architecture; the vision tower, MTP head, and router/embedding/LM-head were
+excluded.
+
+## How to use
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_id = "FLs-AI/FL-9B-4"  # adjust to your repo
+tok = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="bfloat16", device_map="auto")
+
+messages = [{"role": "user", "content": "Translate this COBOL program to Java:\n\n<COBOL here>"}]
+inputs = tok.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
+out = model.generate(inputs, max_new_tokens=2048, temperature=0.0)
+print(tok.decode(out[0][inputs.shape[1]:], skip_special_tokens=True))
+```
+
+## License
+
+Inherits the license of the base model (Qwen/Qwen3.5-9B-Base). Review the base
+model's license terms before use.